# User-Meeting PyRolL 2024

Date: 30th August 2024
Time: 08:00 - 12:00
Participants:

- G. Rothenbucher (GR)
- M. Molter (MM)
- C. Fasching (CF)
- M. Liskow (ML)
- C. Schäfer (CS)
- M. Schmidtchen (MSc)
- M. Weiner (MW)
- M. Stirl (MS)
- J. Mantel (JM)
- K. Hasan (KH)

## Introduction

- Several new Packages, first funded Project with PyRolL, ...

- **Question CR** In several Thesis at our Institute we combined several Packages with PyRolL - Like PyCalphad in Thesis by W. Tell (CCT-Diagrams, ...):
- MM: depends on usage of PyRolL - so depends on who you ask. R&D departments may want that
- CR: several Companies have their own material Databases. Should we invest time to support that or to give examples:
- MSc: We cannot give out all datasets - only from publicly funded
- GR: contact JMatPro representative to get to know what needs to be supported - JMatPro has example Datasets and we should consider to have the ability to support those
- but we shouldnt need to necessarily support all Materials or Invest too much Time into what goes beyond that
- https://matplus.eu/de/
- CR: What about other Packages like MatCalc or Thermocalc?
- Matcalc better as the contact is better - and they want to work with us
- We should rather focus on supporting that and giving examples - what goes beyond is in the responsibility of the user

- **Question CR** should we continue to give examples like the recent velocity calculation in Jupyter Notebooks?:
- Talk about interstand tensions and their relevancy for rolling (came to that topic as we recently created a notebook for that)
- CR: We could provide a notebook that shows an example for calculating interstand tension and adjusting stand RPMs depending on changes to the process
- Also we require the Pillar Model to be accurate for this:
- Currently we have no actually working spreading model for the pillar model (except for maybe Geuze) as they are all one-step models
- MSc: some publication got their stress distributions from FEMs

## Master Thesis KH

- 5-pass ThreeRollPass system is to be optimised using genetic algorithms (SciPy)
- optimisation regarding the inscribed circle diameter
- --> see jupyter notebook

## Example Notebook MM & GR

- In recent meetings (last Kalibreurstag 2023, Kick-Off Validation funding) several shortcomings of the current PyRolL version came up.
- Shortcomings:
- Python based (hard to get into)
- html export not optimised for use in companies
- SI units and not the typical units (mm, °C, ...)
- Notebook:
- basic idea: only one cell to edit, rest is calculating "by itself" without the need for input from user
- additional Packages or usages:
- python-pptx
- import from an excel (which contains roll/ groove contours)
- Input:
- groove list
- specifies roll or groove label for input from the excel
- excel can also diferentiate between Two-/ ThreeRollPasses
- material
- specified in seperate file
- PyRolL should support import of several datatypes to import material parameters im companies
- hot/ cold measure and thermal expansion factor
-
- MM: should more input parameters be already present in the excel table?
- MW: not a problem if more is in the excel sheet, as all values can be changed later on anyways if one wants to test different setups
- Conversion file
- converts non SI units to SI (or the other way)
- GR: Maybe create a wrapper function for all the in profiles? so one doesnt need to support all in-profile types in the notebook.
- MW: we currently do it this way so that ther is Tab-completion but a wrapper function could be possible
- import from dataframe to PyRolL pass sequence structure using PyRolL-from-dict package
- MM: Maybe give possibility to supply different friction factors for different stands?
- difference between carbide rolls or cast iron rolls
- Transport lengths in array
- GR possibility for the customer to provide Meta-information like:
- define Drivetrain
- Stand layout and distances
- MW: we thought about giving a package for drivetrain which can take gearratios and others and then output efficiencies
- or to give errors if my powers or forces exceed the capabilities of my machinery
- tension and velocity calculations
- notebook assumes no tension
- export to dataframe and now one can do with it whatever they want
- conversion back from SI units
- plotting, pandas analysis functions
- modify_df function:
- optimises output values for later use
- horizontal/ vertical orientation
- rename inscribed circle diameter to DSA
- 3D plot of entire sequence
- currently matplotlib
- maybe use plotly?!
- plot to an animation with one frame for each pass
- export to pptx:
- export to tables
- easy to setup plots
- several subplots (boxplots) for typical rolling/ forming parameters
- comparison of actual plant values to simulated ones
- the roll working diameter PyRolL calculates is very good (its just the centre of gravity of the roll contour substractes from the nominal diameter)
- comparison of values from a chinese paper to PyRolL
- also plots contour in cylinder-coordinates
- maybe PyRolL should support that conversion?
- GR: Hill model with additional factor gives the best fit
- MM:
- in ThreeRollPasses Marini spreading model is sometimes significantly better than Wusatowski
- but in TwoRollPasses and some ThreeRollPasses is minimal again
- GR suggestion: Provide contact contour as Polygon, as one currently has to use the zouhar factors to plot it
- there is no global scaling currently
- so each roll pass plot scales scales depending on itself and passes furhter back in the pass sequence get larger and larger
- maybe give ability to provide a global scaling factor depending on the largest roll pass so that they are easily comparable

## MM comparison of examples to different simulation softwares

- comparison of Q-Kaliber, Q-Form, MPC, PyRolL
- for PassSequence of MM
- winconp
- what values should also be given by pyroll
- (tonnage per hour, stand elasticity, productivity values)

## MW assymetric rolling

- probably in PyRolL 3.0 or 2.2 we will implement support for different rolls (for example: different upper/ lower roll in TwoRollPasses)
- needed entirely different structure in base PyRolL code --> therefore new version

## End discussion

- Wishes for development 2025:
- GR: **RPM calculation** from profile velocities and **interstand tension**
- CR: velocity differences previous - next stand as a r-factor (give to pyroll)
- pillar model:
- not too necessary, especially for three-roll-pass
- but it gives real profile contour
- or realise the profile contour with another model
- Future of Pyroll:
- discussion on how to best proceed since next year MW, CR will have little time for Pyroll